Skip to content

Latest commit

 

History

History
68 lines (66 loc) · 5.2 KB

metrics.md

File metadata and controls

68 lines (66 loc) · 5.2 KB

Metrics

OVN-Kubernetes master

This includes a description of a selective set of metrics and to explore the exhausted set, see go-controller/pkg/metrics/master.go

Configuration duration recorder

Setup

Enabled by default with the kind.sh (in directory $ROOT/contrib) Kind setup script. Disabled by default for binary ovnkube-master and enabled with flag --metrics-enable-config-duration.

High-level description

This set of metrics gives a result for the upper bound duration which means, it has taken at most this amount of seconds to apply the configuration to all nodes. It does not represent the exact accurate time to apply only this configuration. Measurement accuracy can be impacted by other parallel processing that might be occurring while the measurement is in progress therefore, the accuracy of the measurements should only indicate upper bound duration to roll out configuration changes.

Metrics

Name Prometheus type Description
ovnkube_master_network_programming_duration_seconds Histogram The duration to apply network configuration for a kind (e.g. pod, service, networkpolicy). Configuration includes add, update and delete events for kinds. This includes OVN-Kubernetes master and OVN duration.
ovnkube_master_network_programming_ovn_duration_seconds Histogram The duration for OVN to apply network configuration for a kind (e.g. pod, service, networkpolicy).

Change log

This list is to help notify if there are additions, changes or removals to metrics. Latest changes are at the top of this list.

  • Effect of OVN IC architecture:
    • Move all the metrics from subsystem "ovnkube-master" to subsystem "ovnkube-controller". The non-IC and IC deployments will each continue to have their ovnkube-master and ovnkube-controller containers running inside the ovnkube-master and ovnkube-controller pods. The metrics scraping should work seemlessly. See ovn-kubernetes#3723 for details
    • Move the following metrics from subsystem "master" to subsystem "clustermanager". Therefore, the follow metrics are renamed.
      • ovnkube_master_num_v4_host_subnets -> ovnkube_clustermanager_num_v4_host_subnets
      • ovnkube_master_num_v6_host_subnets -> ovnkube_clustermanager_num_v6_host_subnets
      • ovnkube_master_allocated_v4_host_subnets -> ovnkube_clustermanager_allocated_v4_host_subnets
      • ovnkube_master_allocated_v6_host_subnets -> ovnkube_clustermanager_allocated_v6_host_subnets
      • ovnkube_master_num_egress_ips -> ovnkube_clustermanager_num_egress_ips
      • ovnkube_master_egress_ips_node_unreachable_total -> ovnkube_clustermanager_egress_ips_node_unreachable_total
      • ovnkube_master_egress_ips_rebalance_total -> ovnkube_clustermanager_egress_ips_rebalance_total
  • Update description of ovnkube_master_pod_creation_latency_seconds
  • Add libovsdb metrics - ovnkube_master_libovsdb_disconnects_total and ovnkube_master_libovsdb_monitors.
  • Add ovn_controller_southbound_database_connected metric (ovn-kubernetes#3117).
  • Stopwatch metrics now report in seconds instead of milliseconds.
  • Rename (ovn-kubernetes#3022):
    • ovs_vswitchd_interface_link_resets -> ovs_vswitchd_interface_resets_total
    • ovs_vswitchd_interface_rx_dropped -> ovs_vswitchd_interface_rx_dropped_total
    • ovs_vswitchd_interface_tx_dropped -> ovs_vswitchd_interface_tx_dropped_total
    • ovs_vswitchd_interface_rx_errors -> ovs_vswitchd_interface_rx_errors_total
    • ovs_vswitchd_interface_tx_errors -> ovs_vswitchd_interface_tx_errors_total
    • ovs_vswitchd_interface_collisions -> ovs_vswitchd_interface_collisions_total
  • Remove (ovn-kubernetes#3022):
    • ovs_vswitchd_dp_if
    • ovs_vswitchd_interface_driver_name
    • ovs_vswitchd_interface_driver_version
    • ovs_vswitchd_interface_firmware_version
    • ovs_vswitchd_interface_rx_packets
    • ovs_vswitchd_interface_tx_packets
    • ovs_vswitchd_interface_rx_bytes
    • ovs_vswitchd_interface_tx_bytes
    • ovs_vswitchd_interface_rx_frame_err
    • ovs_vswitchd_interface_rx_over_err
    • ovs_vswitchd_interface_rx_crc_err
    • ovs_vswitchd_interface_name
    • ovs_vswitchd_interface_duplex
    • ovs_vswitchd_interface_type
    • ovs_vswitchd_interface_admin_state
    • ovs_vswitchd_interface_link_state
    • ovs_vswitchd_interface_ifindex
    • ovs_vswitchd_interface_link_speed
    • ovs_vswitchd_interface_mtu
    • ovs_vswitchd_interface_ofport
    • ovs_vswitchd_interface_ingress_policing_burst
    • ovs_vswitchd_interface_ingress_policing_rate
  • Add ovnkube_master_network_programming_duration_seconds and ovnkube_master_network_programming_ovn_duration_seconds (ovn-kubernetes#2878)
  • Remove ovnkube_master_skipped_nbctl_daemon_total (ovn-kubernetes#2707)
  • Add ovnkube_master_egress_routing_via_host (ovn-kubernetes#2833)
  • Add ovnkube_resource_retry_failures_total (ovn-kubernetes#3314)
  • Add ovs_vswitchd_interfaces_total and ovs_vswitchd_interface_up_wait_seconds_total (ovn-kubernetes#3391)